06-pytorch-transfer-learning: got some error in [3.2 Setting up a pretrained model] (torch:2.1.0+cu121, torchvision:0.16.0+cu121)

[stleafz]

I followed the video("Setting Up a Pretrained Model with Torchvision") and tried to run the code :

model = torchvision.models.efficientnet_b0(weights=weights)
model

then got the error message:
invalid hash value (expected "3dd342df", got "7f5810bc96def8f7552d5b7e68d53c4786f81167d28291b21c0d90e1fca14934")

It seems like a new bug, and it could be fixed temporarily by overriding torchvision.models._api.WeightsEnum.get_state_dict()
source:pytorch/vision#7744 (comment)
I hope this could be helpful to those who got the same error.

[avishekiitism]

thanks

[mrdbourke]

Thank you for the heads up!

A few other people are getting that too.

See the following issue/comment for further code/solution: #696 (comment)

[mustious]

If you still face the same issue, after running the first code cell to install the packages, try to Restart Session on Google Colab. Then, run the imports again.

The issue seems to be with torchvision v0.16.0 specifically. Other higher versions work seamlessly. pytorch/vision#7744 (comment)